mirror of
https://github.com/NVIDIA-NeMo/DataDesigner
synced 2026-05-24 09:48:29 +00:00
The local counter `curr_num_correction_steps` in `ModelFacade.generate` and `agenerate` was incremented before each parse attempt, so it actually counted parse attempts (initial + corrections), not corrections taken. The name suggested the latter, which made the surrounding `<= max_correction_steps` check read like a possible off-by-one even though the math worked out. Rename it to `parse_attempts` and add a short comment describing the semantics. Pure refactor: no behavior change. Closes #371 |
||
|---|---|---|
| .. | ||
| src/data_designer/engine | ||
| tests | ||
| pyproject.toml | ||
| README.md | ||
data-designer-engine
Generation engine for NeMo Data Designer synthetic data generation framework.
This package contains the execution engine that powers Data Designer. It depends on data-designer-config and includes heavy dependencies like pandas, numpy, and native HTTP-based LLM integration.
Installation
pip install data-designer-engine
This automatically installs data-designer-config as a dependency.
See main README.md for more information.